Raw model output on the native 0.25° grid — no MOS, no bias correction, no post-processing. Scores understate operational forecast quality. Read the fairness statement →

Station

KSEA — Seattle-Tacoma

47.4447, -122.3136 · 130 m · standard offset UTC-8 h · truth product CLISEA

Window: the last 90 days
Init: 00Z
Interpolation: bilinear
Data through: 2026-10-28
Updated: 2026-10-28 17:07 UTC
Next update: 2026-10-29 11:00 UTC

Instantaneous temperature — MAE °Ft2

Scored against the observed 2 m temperature at the same instant (ASOS routine METAR). This is the headline variable: the same four instants on the forecast and the observation side. These pages are always bilinear interpolation, the headline method. The nearest-node variant, and every other window/initialization/interpolation/variable combination, is published on each permanent link below — follow any number.

the last 90 days, 00Z, bilinear. MAE with the bias and n underneath; every cell links to its permanent page. The sparkline shows the same row across lead days 1–9 on a scale shared within this table.
Model	d0	d1	d2	d3	d4	d5	d6	d7	d8	d9
ECMWF AIFS Singleaifs_single	3.00−2.68 · n=88	2.72−2.21 · n=87	2.57−2.02 · n=86	2.57−1.92 · n=85	2.51−1.52 · n=84	2.55−1.33 · n=83	2.91−1.54 · n=82	3.28−1.76 · n=81	3.50−1.78 · n=80	—
Aurora (GFS init)aurora_gfs	2.08−0.59 · n=56	2.21−1.17 · n=56	2.46−1.26 · n=55	2.69−1.26 · n=54	2.85−1.57 · n=55	3.10−1.55 · n=56	3.21−1.09 · n=55	3.42−1.01 · n=55	3.55−1.05 · n=55	—
Aurora (IFS init)aurora_ifs	2.11−0.90 · n=82	2.20−0.85 · n=82	2.19−0.78 · n=82	2.28−0.68 · n=82	2.39−0.71 · n=81	2.71−0.30 · n=81	3.05−0.03 · n=81	3.38+0.27 · n=81	3.63+0.76 · n=81	—
FourCastNet v2 (GFS init)fourcastnet_gfs	2.94+0.69 · n=56	2.81+0.70 · n=56	3.11+1.17 · n=55	3.09+1.32 · n=54	3.42+0.83 · n=55	3.41+0.27 · n=56	3.65+0.49 · n=55	4.20+0.42 · n=55	4.37+0.39 · n=55	—
FourCastNet v2 (IFS init)fourcastnet_ifs	2.73+0.44 · n=87	2.83+1.05 · n=87	3.05+1.32 · n=87	3.15+1.47 · n=87	3.30+1.30 · n=87	3.51+0.97 · n=87	3.72+0.79 · n=87	4.25+0.56 · n=87	4.41+0.76 · n=87	—
NCEP GFSgfs	3.34+0.45 · n=88	3.35+0.42 · n=87	3.46+0.21 · n=86	3.57+0.35 · n=85	3.79+0.39 · n=84	4.06−0.36 · n=83	4.35+0.32 · n=82	4.63+0.14 · n=81	4.93+0.44 · n=80	—
GraphCast (GFS init)graphcast_gfs	2.46−0.76 · n=56	2.46−1.39 · n=56	2.60−1.41 · n=55	2.71−1.26 · n=54	2.80−1.21 · n=55	3.08−1.38 · n=56	3.20−1.07 · n=55	3.73−1.08 · n=55	3.73−0.97 · n=55	—
GraphCast (IFS init)graphcast_ifs	2.15−0.80 · n=86	2.30−0.97 · n=86	2.35−1.09 · n=86	2.46−0.94 · n=86	2.62−0.86 · n=86	2.75−0.96 · n=86	3.13−1.05 · n=86	3.37−1.16 · n=86	3.57−0.94 · n=86	—
ECMWF IFS HRESifs_hres	1.95−0.15 · n=88	2.15−0.16 · n=87	2.25−0.18 · n=86	2.51−0.21 · n=85	2.52−0.36 · n=84	2.70−0.38 · n=83	3.12−0.22 · n=82	3.68−0.37 · n=81	3.63−0.62 · n=80	—
Pangu-Weather (GFS init)pangu_gfs	2.19+1.17 · n=56	2.18+0.71 · n=56	2.49+0.68 · n=55	2.69+0.63 · n=54	2.75+0.55 · n=55	3.05+0.35 · n=56	3.46+0.65 · n=55	3.96+0.89 · n=55	4.12+0.54 · n=55	—
Pangu-Weather (IFS init)pangu_ifs	2.07+0.29 · n=84	2.17+0.26 · n=84	2.27+0.25 · n=84	2.38+0.18 · n=84	2.65−0.01 · n=84	2.95−0.26 · n=84	3.36−0.20 · n=84	3.81−0.23 · n=84	4.22+0.02 · n=84	—
Persistence (baseline)persistence	—	3.26+0.16 · n=89	3.99+0.31 · n=89	4.28+0.46 · n=89	4.17+0.68 · n=89	4.41+0.84 · n=89	4.33+0.98 · n=89	4.32+1.14 · n=89	4.26+1.29 · n=89	4.68+1.57 · n=89

+ model too warm − model too cold bias interval includes zero ★ lowest MAE among ranked models (n ≥ 30) = not distinguishable from the leader ▼ worse than the leader ▲ better than the leader — all three Holm-corrected within this table, over the family of comparisons against the leader; the uncorrected verdict for every pair is published in the pairwise table of each permanent link and in pairwise_latest.csv.gzn < 30 greyed and unranked CI is a 95 % moving-block bootstrap interval on the group's own days, and reads — when it could not be computed skill reads — when fewer than 10 days are common to the model and the baseline: the ratio of two means over a handful of shared days is not a number worth printing

Sampled daily maximum — MAE °Ftmax_s

Scored against the maximum of the four *observed* samples on the same day.

the last 90 days, 00Z, bilinear. MAE with the bias and n underneath; every cell links to its permanent page. The sparkline shows the same row across lead days 1–9 on a scale shared within this table.
Model	d0	d1	d2	d3	d4	d5	d6	d7	d8	d9
ECMWF AIFS Singleaifs_single	2.89−2.70 · n=88	2.73−2.38 · n=87	2.71−2.34 · n=86	2.91−2.49 · n=85	2.89−2.09 · n=84	3.11−2.10 · n=83	3.56−2.41 · n=82	4.09−2.75 · n=81	4.12−2.41 · n=80	—
Aurora (GFS init)aurora_gfs	2.72−2.38 · n=56	3.20−2.93 · n=56	3.41−3.12 · n=55	4.02−3.44 · n=54	4.15−3.52 · n=55	4.49−3.27 · n=56	4.08−2.63 · n=55	4.57−2.76 · n=55	5.08−2.92 · n=55	—
Aurora (IFS init)aurora_ifs	1.91−1.37 · n=82	2.17−1.30 · n=82	2.23−1.26 · n=82	2.44−1.27 · n=82	2.60−1.29 · n=81	3.03−0.92 · n=81	3.52−0.67 · n=81	4.03−0.18 · n=81	4.35+0.58 · n=81	—
FourCastNet v2 (GFS init)fourcastnet_gfs	2.35+0.94 · n=56	2.80+0.98 · n=56	2.93+1.61 · n=55	2.80+1.56 · n=54	3.46+1.33 · n=55	3.74+0.65 · n=56	4.03+1.01 · n=55	5.33+0.57 · n=55	5.33+0.77 · n=55	—
FourCastNet v2 (IFS init)fourcastnet_ifs	2.02+0.77 · n=87	2.39+1.32 · n=87	2.73+1.68 · n=87	3.05+1.74 · n=87	3.34+1.58 · n=87	3.92+1.17 · n=87	4.46+1.00 · n=87	5.44+0.63 · n=87	5.33+1.20 · n=87	—
NCEP GFSgfs	3.49+2.21 · n=88	3.47+2.29 · n=87	3.57+1.94 · n=86	3.65+2.44 · n=85	4.12+2.49 · n=84	4.58+1.09 · n=83	4.74+2.46 · n=82	5.18+2.04 · n=81	6.51+2.42 · n=80	—
GraphCast (GFS init)graphcast_gfs	3.58−3.31 · n=56	3.73−3.54 · n=56	3.89−3.61 · n=55	3.95−3.58 · n=54	3.89−2.95 · n=55	4.49−3.26 · n=56	4.53−2.61 · n=55	5.14−3.09 · n=55	5.75−2.85 · n=55	—
GraphCast (IFS init)graphcast_ifs	2.75−2.56 · n=86	2.92−2.62 · n=86	3.07−2.78 · n=86	3.08−2.54 · n=86	3.23−2.46 · n=86	3.40−2.73 · n=86	4.14−2.81 · n=86	4.51−2.76 · n=86	4.81−2.14 · n=86	—
ECMWF IFS HRESifs_hres	1.88−0.36 · n=88	2.24−0.34 · n=87	2.39−0.39 · n=86	2.61−0.46 · n=85	2.77−0.92 · n=84	3.22−0.84 · n=83	3.78−0.31 · n=82	4.38−0.66 · n=81	4.52−0.77 · n=80	—
Pangu-Weather (GFS init)pangu_gfs	1.97+0.85 · n=56	2.17−0.22 · n=56	2.20−0.34 · n=55	2.62−1.07 · n=54	2.69−0.88 · n=55	3.58−1.17 · n=56	3.90−0.78 · n=55	4.71−0.73 · n=55	4.65−1.12 · n=55	—
Pangu-Weather (IFS init)pangu_ifs	1.75−0.54 · n=84	1.96−0.61 · n=84	2.17−0.74 · n=84	2.37−1.03 · n=84	2.78−1.24 · n=84	3.27−1.81 · n=84	4.10−1.54 · n=84	4.87−1.70 · n=84	5.59−1.26 · n=84	—
Persistence (baseline)persistence	—	3.36+0.26 · n=89	4.77+0.47 · n=89	5.51+0.59 · n=89	5.44+0.85 · n=89	5.47+1.00 · n=89	5.63+1.18 · n=89	5.47+1.38 · n=89	5.54+1.54 · n=89	6.04+1.88 · n=89

Sampled daily minimum — MAE °Ftmin_s

Scored against the minimum of the four *observed* samples on the same day.

the last 90 days, 00Z, bilinear. MAE with the bias and n underneath; every cell links to its permanent page. The sparkline shows the same row across lead days 1–9 on a scale shared within this table.
Model	d0	d1	d2	d3	d4	d5	d6	d7	d8	d9
ECMWF AIFS Singleaifs_single	3.28−3.21 · n=88	2.62−2.45 · n=87	2.32−2.11 · n=86	2.04−1.76 · n=85	1.87−1.39 · n=84	1.88−1.06 · n=83	2.05−0.98 · n=82	2.23−0.95 · n=81	2.80−1.32 · n=80	—
Aurora (GFS init)aurora_gfs	1.48+0.81 · n=56	1.39+0.39 · n=56	1.57+0.25 · n=55	1.68+0.49 · n=54	1.73−0.03 · n=55	1.98−0.22 · n=56	2.44+0.17 · n=55	2.23+0.37 · n=55	2.45+0.57 · n=55	—
Aurora (IFS init)aurora_ifs	1.67−0.75 · n=82	1.58−0.70 · n=82	1.49−0.68 · n=82	1.48−0.48 · n=82	1.48−0.41 · n=81	1.78−0.10 · n=81	2.13+0.21 · n=81	2.12+0.27 · n=81	2.41+0.66 · n=81	—
FourCastNet v2 (GFS init)fourcastnet_gfs	2.61−1.12 · n=56	1.95−0.89 · n=56	1.98−0.66 · n=55	2.07−0.30 · n=54	2.36−0.90 · n=55	2.46−1.40 · n=56	2.76−1.21 · n=55	2.83−1.18 · n=55	2.96−1.16 · n=55	—
FourCastNet v2 (IFS init)fourcastnet_ifs	2.49−1.51 · n=87	1.94−0.66 · n=87	1.99−0.57 · n=87	1.88−0.22 · n=87	2.05−0.32 · n=87	2.22−0.62 · n=87	2.25−0.77 · n=87	2.60−0.87 · n=87	2.91−1.06 · n=87	—
NCEP GFSgfs	2.06−1.50 · n=88	2.14−1.69 · n=87	2.21−1.74 · n=86	2.32−1.73 · n=85	2.49−2.05 · n=84	2.62−2.13 · n=83	3.20−2.26 · n=82	3.35−2.07 · n=81	3.14−2.00 · n=80	—
GraphCast (GFS init)graphcast_gfs	1.77+1.23 · n=56	1.50+0.21 · n=56	1.47+0.11 · n=55	1.64+0.37 · n=54	1.75−0.07 · n=55	2.07−0.23 · n=56	2.07−0.12 · n=55	2.21+0.16 · n=55	2.57+0.34 · n=55	—
GraphCast (IFS init)graphcast_ifs	1.33+0.24 · n=86	1.44−0.01 · n=86	1.45−0.16 · n=86	1.50−0.04 · n=86	1.66+0.02 · n=86	1.66−0.00 · n=86	2.00−0.04 · n=86	2.16−0.19 · n=86	2.35−0.37 · n=86	—
ECMWF IFS HRESifs_hres	1.52−0.49 · n=88	1.54−0.42 · n=87	1.50−0.39 · n=86	1.70−0.39 · n=85	1.85−0.25 · n=84	1.97−0.54 · n=83	2.05−0.50 · n=82	2.63−0.59 · n=81	2.69−0.84 · n=80	—
Pangu-Weather (GFS init)pangu_gfs	1.94+1.15 · n=56	1.60+0.87 · n=56	1.95+0.78 · n=55	2.27+1.00 · n=54	1.81+0.67 · n=55	2.12+0.45 · n=56	2.59+0.63 · n=55	2.25+0.89 · n=55	3.09+0.66 · n=55	—
Pangu-Weather (IFS init)pangu_ifs	1.82+0.19 · n=84	1.73+0.22 · n=84	1.77+0.21 · n=84	1.82+0.26 · n=84	1.86+0.21 · n=84	2.15+0.01 · n=84	2.19−0.15 · n=84	2.56−0.15 · n=84	3.10+0.05 · n=84	—
Persistence (baseline)persistence	—	2.37+0.08 · n=89	2.73+0.19 · n=89	3.05+0.35 · n=89	2.97+0.54 · n=89	3.05+0.69 · n=89	3.00+0.80 · n=89	2.99+0.92 · n=89	3.16+1.07 · n=89	3.27+1.31 · n=89

Against the NWS daily extremes (secondary)

This comparison scores four 6-hourly forecast samples against the true daily extreme, so it includes a sampling penalty. The size of that penalty depends on each model's own diurnal amplitude, which is one of the things this site measures, so these numbers are published for operational relevance and are never used for ranking.

Daily maximum vs NWS CLI

the last 90 days, 00Z, bilinear. MAE with the bias and n underneath; every cell links to its permanent page. The sparkline shows the same row across lead days 1–9 on a scale shared within this table.
Model	d0	d1	d2	d3	d4	d5	d6	d7	d8	d9
ECMWF AIFS Singleaifs_single	4.16−4.11 · n=88	3.93−3.79 · n=87	3.88−3.74 · n=86	4.01−3.89 · n=85	3.96−3.49 · n=84	3.91−3.51 · n=83	4.47−3.82 · n=82	4.97−4.17 · n=81	5.00−3.84 · n=80	—
Aurora (GFS init)aurora_gfs	3.93−3.76 · n=56	4.57−4.48 · n=56	4.72−4.57 · n=55	5.11−4.90 · n=54	5.29−4.86 · n=55	5.30−4.73 · n=56	4.86−4.13 · n=55	5.20−4.22 · n=55	5.70−4.28 · n=55	—
Aurora (IFS init)aurora_ifs	2.93−2.78 · n=82	3.07−2.71 · n=82	3.18−2.67 · n=82	3.26−2.63 · n=82	3.35−2.74 · n=81	3.56−2.27 · n=81	3.80−2.03 · n=81	4.22−1.57 · n=81	4.40−0.90 · n=81	—
FourCastNet v2 (GFS init)fourcastnet_gfs	2.32−0.44 · n=56	3.07−0.57 · n=56	3.01+0.15 · n=55	2.86+0.09 · n=54	3.59−0.01 · n=55	3.85−0.81 · n=56	4.15−0.50 · n=55	5.43−0.88 · n=55	5.39−0.59 · n=55	—
FourCastNet v2 (IFS init)fourcastnet_ifs	2.13−0.65 · n=87	2.39−0.10 · n=87	2.69+0.26 · n=87	3.05+0.36 · n=87	3.47+0.16 · n=87	4.11−0.21 · n=87	4.72−0.42 · n=87	5.82−0.80 · n=87	5.64−0.21 · n=87	—
NCEP GFSgfs	3.17+0.81 · n=88	3.35+0.88 · n=87	3.52+0.53 · n=86	3.33+1.03 · n=85	3.89+1.09 · n=84	4.52−0.32 · n=83	4.59+1.05 · n=82	4.98+0.62 · n=81	5.91+0.99 · n=80	—
GraphCast (GFS init)graphcast_gfs	4.82−4.70 · n=56	5.16−5.09 · n=56	5.24−5.07 · n=55	5.25−5.04 · n=54	5.01−4.28 · n=55	5.43−4.71 · n=56	5.56−4.11 · n=55	5.92−4.54 · n=55	6.14−4.21 · n=55	—
GraphCast (IFS init)graphcast_ifs	4.01−3.96 · n=86	4.17−4.03 · n=86	4.35−4.18 · n=86	4.23−3.94 · n=86	4.34−3.84 · n=86	4.41−4.11 · n=86	5.14−4.24 · n=86	5.21−4.15 · n=86	5.33−3.56 · n=86	—
ECMWF IFS HRESifs_hres	2.37−1.76 · n=88	2.67−1.75 · n=87	2.84−1.79 · n=86	2.95−1.87 · n=85	3.43−2.32 · n=84	3.70−2.24 · n=83	4.06−1.72 · n=82	4.55−2.07 · n=81	4.98−2.20 · n=80	—
Pangu-Weather (GFS init)pangu_gfs	2.07−0.54 · n=56	2.84−1.76 · n=56	3.01−1.79 · n=55	3.31−2.54 · n=54	3.25−2.22 · n=55	4.08−2.63 · n=56	4.56−2.28 · n=55	5.11−2.18 · n=55	5.11−2.48 · n=55	—
Pangu-Weather (IFS init)pangu_ifs	2.41−1.97 · n=84	2.71−2.02 · n=84	2.89−2.16 · n=84	3.00−2.39 · n=84	3.67−2.68 · n=84	3.97−3.14 · n=84	4.85−2.95 · n=84	5.44−3.12 · n=84	6.07−2.71 · n=84	—
Persistence (baseline)persistence	—	3.33+0.25 · n=89	4.66+0.44 · n=89	5.30+0.61 · n=89	5.35+0.88 · n=89	5.39+1.06 · n=89	5.27+1.20 · n=89	5.25+1.40 · n=89	5.40+1.56 · n=89	5.93+1.93 · n=89

Daily minimum vs NWS CLI

the last 90 days, 00Z, bilinear. MAE with the bias and n underneath; every cell links to its permanent page. The sparkline shows the same row across lead days 1–9 on a scale shared within this table.
Model	d0	d1	d2	d3	d4	d5	d6	d7	d8	d9
ECMWF AIFS Singleaifs_single	2.03−1.70 · n=88	1.56−0.94 · n=87	1.41−0.61 · n=86	1.34−0.27 · n=85	1.29+0.11 · n=84	1.49+0.44 · n=83	1.86+0.50 · n=82	2.11+0.52 · n=81	2.54+0.16 · n=80	—
Aurora (GFS init)aurora_gfs	2.33+2.24 · n=56	1.97+1.87 · n=56	1.84+1.72 · n=55	2.13+2.08 · n=54	1.84+1.33 · n=55	1.87+1.41 · n=56	2.51+1.66 · n=55	2.49+1.96 · n=55	2.76+1.93 · n=55	—
Aurora (IFS init)aurora_ifs	1.36+0.77 · n=82	1.25+0.78 · n=82	1.30+0.89 · n=82	1.40+0.98 · n=82	1.49+1.14 · n=81	1.89+1.41 · n=81	2.25+1.74 · n=81	2.37+1.79 · n=81	2.76+2.20 · n=81	—
FourCastNet v2 (GFS init)fourcastnet_gfs	2.12+0.31 · n=56	1.62+0.60 · n=56	1.76+0.82 · n=55	1.94+1.29 · n=54	2.09+0.46 · n=55	2.17+0.23 · n=56	2.41+0.29 · n=55	2.43+0.41 · n=55	2.70+0.21 · n=55	—
FourCastNet v2 (IFS init)fourcastnet_ifs	2.16+0.04 · n=87	1.86+0.85 · n=87	1.81+1.00 · n=87	1.98+1.30 · n=87	1.96+1.23 · n=87	2.12+0.92 · n=87	2.13+0.77 · n=87	2.21+0.67 · n=87	2.57+0.45 · n=87	—
NCEP GFSgfs	1.57+0.02 · n=88	1.60−0.18 · n=87	1.70−0.24 · n=86	1.73−0.25 · n=85	1.95−0.55 · n=84	2.05−0.64 · n=83	2.56−0.78 · n=82	2.79−0.60 · n=81	2.68−0.52 · n=80	—
GraphCast (GFS init)graphcast_gfs	2.73+2.66 · n=56	1.88+1.70 · n=56	1.79+1.59 · n=55	2.09+1.96 · n=54	1.87+1.29 · n=55	2.03+1.40 · n=56	2.33+1.37 · n=55	2.49+1.74 · n=55	2.84+1.71 · n=55	—
GraphCast (IFS init)graphcast_ifs	1.88+1.78 · n=86	1.64+1.50 · n=86	1.61+1.38 · n=86	1.73+1.45 · n=86	1.88+1.56 · n=86	1.91+1.51 · n=86	2.09+1.51 · n=86	2.30+1.37 · n=86	2.39+1.18 · n=86	—
ECMWF IFS HRESifs_hres	1.49+1.02 · n=88	1.55+1.09 · n=87	1.60+1.12 · n=86	1.74+1.10 · n=85	1.94+1.25 · n=84	1.86+0.96 · n=83	1.91+0.98 · n=82	2.28+0.89 · n=81	2.43+0.64 · n=80	—
Pangu-Weather (GFS init)pangu_gfs	2.73+2.58 · n=56	2.42+2.36 · n=56	2.48+2.25 · n=55	2.77+2.59 · n=54	2.34+2.04 · n=55	2.75+2.07 · n=56	3.12+2.12 · n=55	3.04+2.48 · n=55	3.52+2.02 · n=55	—
Pangu-Weather (IFS init)pangu_ifs	2.13+1.74 · n=84	2.03+1.72 · n=84	2.14+1.79 · n=84	2.04+1.70 · n=84	2.07+1.75 · n=84	2.26+1.53 · n=84	2.34+1.38 · n=84	2.77+1.37 · n=84	3.29+1.57 · n=84	—
Persistence (baseline)persistence	—	2.06+0.08 · n=89	2.40+0.16 · n=89	2.63+0.29 · n=89	2.69+0.51 · n=89	2.85+0.67 · n=89	2.69+0.80 · n=89	2.91+0.96 · n=89	2.94+1.12 · n=89	3.09+1.34 · n=89

Native daily maximum vs NWS CLI

the last 90 days, 00Z, bilinear. MAE with the bias and n underneath; every cell links to its permanent page. The sparkline shows the same row across lead days 1–9 on a scale shared within this table.
Model	d0	d1	d2	d3	d4	d5	d6	d7	d8	d9
NCEP GFSgfs	3.54+2.55 · n=88	3.50+2.44 · n=87	3.67+2.18 · n=86	3.79+2.65 · n=85	4.22+2.93 · n=84	4.51+1.29 · n=83	4.90+2.98 · n=82	5.70+2.49 · n=81	6.58+2.88 · n=80	—
ECMWF IFS HRESifs_hres	1.77−0.85 · n=88	2.17−0.80 · n=87	2.31−0.79 · n=86	2.50−0.93 · n=85	2.91−1.38 · n=83	3.07−1.17 · n=83	3.72−0.80 · n=82	4.20−0.99 · n=81	4.59−1.22 · n=80	—
Persistence (baseline)persistence	—	3.31+0.31 · n=88	4.61+0.55 · n=88	5.24+0.74 · n=88	5.35+0.94 · n=88	5.34+1.18 · n=88	5.20+1.34 · n=88	5.28+1.44 · n=88	5.38+1.67 · n=88	5.88+1.83 · n=88

Native daily minimum vs NWS CLI

the last 90 days, 00Z, bilinear. MAE with the bias and n underneath; every cell links to its permanent page. The sparkline shows the same row across lead days 1–9 on a scale shared within this table.
Model	d0	d1	d2	d3	d4	d5	d6	d7	d8	d9
NCEP GFSgfs	2.23−1.66 · n=88	2.15−1.71 · n=87	2.37−1.84 · n=86	2.38−1.95 · n=85	2.54−2.10 · n=84	2.78−2.22 · n=83	3.13−2.35 · n=82	3.38−2.19 · n=81	3.11−2.10 · n=80	—
ECMWF IFS HRESifs_hres	1.16−0.00 · n=88	1.23+0.14 · n=87	1.36+0.17 · n=86	1.41+0.25 · n=85	1.67+0.18 · n=83	1.85−0.45 · n=83	1.98−0.49 · n=82	2.25−0.72 · n=81	2.63−0.90 · n=80	—
Persistence (baseline)persistence	—	2.06+0.10 · n=88	2.43+0.16 · n=88	2.64+0.27 · n=88	2.68+0.48 · n=88	2.85+0.65 · n=88	2.70+0.80 · n=88	2.93+0.95 · n=88	2.92+1.08 · n=88	3.02+1.25 · n=88

Month by month

MAE °F at lead day 1, 00Z, bilinear, instantaneous temperature — the seasonal slice behind the window averages. A model that is excellent in one season and poor in another shows up here and nowhere else.

Month	aifs_single	aurora_gfs	aurora_ifs	fourcastnet_gfs	fourcastnet_ifs	gfs	graphcast_gfs	graphcast_ifs	ifs_hres	pangu_gfs	pangu_ifs
2026-10	2.49-1.84 · n=109	2.29-1.49 · n=68	2.22-1.34 · n=97	2.66+0.20 · n=68	2.56+0.29 · n=109	2.46+0.39 · n=109	2.36-1.23 · n=68	2.09-0.96 · n=105	1.91-0.86 · n=109	2.23+0.39 · n=68	2.05+0.07 · n=101
2026-09	2.52-2.16 · n=120	2.03-1.25 · n=80	1.88-0.80 · n=104	2.78+0.34 · n=80	2.56+0.78 · n=112	2.83-0.56 · n=120	2.29-1.64 · n=80	2.14-1.25 · n=116	2.05-0.74 · n=120	1.93+0.46 · n=80	1.91+0.29 · n=108
2026-08	3.14-2.60 · n=120	2.31-0.79 · n=76	2.47-0.52 · n=124	2.97+1.53 · n=76	3.29+1.97 · n=124	4.68+1.40 · n=120	2.74-1.25 · n=76	2.63-0.72 · n=120	2.48+1.03 · n=120	2.41+1.25 · n=76	2.53+0.39 · n=124
2026-07	—	2.79-2.00 · n=48	2.32-1.55 · n=80	2.53+0.03 · n=48	2.93+1.36 · n=80	—	2.56-1.35 · n=48	2.32-1.26 · n=124	—	2.45-0.24 · n=48	2.19-0.06 · n=80
2026-06	—	2.37-2.21 · n=64	1.76-1.43 · n=12	2.21+0.19 · n=64	3.10+1.93 · n=12	—	2.33-1.97 · n=64	2.59-1.71 · n=48	—	2.17-1.00 · n=64	1.79-0.82 · n=12
2026-05	—	2.49-1.20 · n=75	2.61-1.08 · n=40	3.43+1.74 · n=75	3.60+2.26 · n=40	—	2.53-1.16 · n=75	2.38-1.01 · n=40	—	2.92+0.26 · n=75	2.73+0.02 · n=40
2026-04	—	2.55-2.09 · n=96	2.31-1.52 · n=116	2.38-0.05 · n=96	2.52+0.56 · n=116	—	3.22-2.90 · n=96	2.77-2.09 · n=116	—	2.27-0.95 · n=96	2.21-0.93 · n=116
2026-03	—	1.96-0.85 · n=124	1.90-0.61 · n=124	2.45-1.11 · n=124	2.40-0.85 · n=124	—	2.07-1.18 · n=124	1.83-0.70 · n=124	—	2.05+1.05 · n=124	1.83+0.68 · n=124
2026-02	—	2.19-1.54 · n=112	2.09-1.72 · n=112	2.82-2.33 · n=112	2.59-2.02 · n=112	—	2.02-1.44 · n=112	1.84-1.29 · n=112	—	2.03-0.20 · n=112	1.91-0.89 · n=112
2026-01	—	2.34-1.38 · n=120	2.52-2.14 · n=120	3.22-2.80 · n=120	3.54-3.05 · n=120	—	2.23-1.66 · n=120	2.23-1.70 · n=120	—	2.20-0.21 · n=120	2.22-1.25 · n=120

Model availability at this station

Each model is scored only over its own available period, on a shared scale; the window averages above are therefore not identical across models.

Model	Period	Scored days
ECMWF AIFS Single	2026-08-02 → 2026-10-27	87
Aurora (GFS init)	2026-01-02 → 2026-10-27	215
Aurora (IFS init)	2026-01-02 → 2026-10-27	232
FourCastNet v2 (GFS init)	2026-01-02 → 2026-10-27	215
FourCastNet v2 (IFS init)	2026-01-02 → 2026-10-27	237
NCEP GFS	2026-08-02 → 2026-10-27	87
GraphCast (GFS init)	2026-01-02 → 2026-10-27	215
GraphCast (IFS init)	2026-01-02 → 2026-10-27	256
ECMWF IFS HRES	2026-08-02 → 2026-10-27	87
Pangu-Weather (GFS init)	2026-01-02 → 2026-10-27	215
Pangu-Weather (IFS init)	2026-01-02 → 2026-10-27	234
Persistence (baseline)	2026-01-01 → 2026-10-27	298

Observed truthlast 30 days

Climatological day	Tmax °F	Tmin °F	Source	Flag
2026-10-27	57	47	CLI
2026-10-26	58	49	CLI
2026-10-25	58	46	CLI
2026-10-24	55	40	CLI
2026-10-23	58	44	CLI
2026-10-22	60	46	CLI	obs_diff_gt2f
2026-10-21	64	43	CLI
2026-10-20	62	46	CLI
2026-10-19	62	46	CLI	obs_diff_gt2f
2026-10-18	60	43	CLI
2026-10-17	62	49	CLI
2026-10-16	77	46	CLI
2026-10-15	68	50	CLI
2026-10-14	67	51	CLI	obs_diff_gt2f
2026-10-13	62	46	CLI
2026-10-12	61	41	CLI
2026-10-11	60	43	CLI
2026-10-10	58	45	CLI
2026-10-09	58	50	CLI
2026-10-08	67	51	CLI
2026-10-07	67	53	CLI
2026-10-06	66	52	CLI
2026-10-05	68	52	CLI
2026-10-04	69	51	CLI
2026-10-03	67	54	CLI
2026-10-02	71	52	CLI
2026-10-01	68	51	CLI
2026-09-30	65	49	CLI
2026-09-29	63	53	CLI
2026-09-28	64	46	CLI